Signal generator creates input stage oscillation? What's up?

[jamesmafyew]

I posted in a previous thread about this problem, but what I found warranted a separate topic.

http://music-electronics-forum.com/t27307-2/#post252981

Above is where I was asking about plate-to-grid caps on the input stage of an amp I built. I had a situation similar to Chuck H where I had the jacks close to the socket, so I just ran the 68k grid stopper resistor leads from the jack to the socket. No shielded wire. With all knobs at 10 and a cable plugged in (but under no other conditions), the amp would oscillate at 40kHz on the scope. Turn any knob down and the oscillation goes away. Unplug the cable and the oscillation goes away. The amp never squealed during normal usage.

So I added various caps from grid to plate, no change. Even with big ones (10pf!). I added shielded wire for the inputs and NFB loop. This whole time, the cable I have plugged in is the cable to my old Heathkit 5218. I tried ungrounding the cable to break the ground loop, the oscillation got worse. However, I have the square wave output hooked up to the trigger input on my scope, for convenience. Unplugged THAT, and the oscillation is gone, no matter what I do. Amp is solid as a rock.

Yet there are still ground loops here -- through the wall socket ground, through the sine output cable. Why the square wave output to the trigger input? Can someone explain this to me? I've never run into this before. I normally just have the scope channel 1 hooked up across my dummy load, the heathkit square wave output plugged into the trigger input, and the sine wave output into the front of the amp. It's never really given me problems before.

(Regardless, after all that, I'm keeping the shielded wire and 1pf grid to plate caps. No compromise to the already-ideal tone and if it makes the amp more stable, great)

[lowell]

I'm confused by your post. Do you have the amp connected to a speaker? Is the oscillation audible? Or are you just seeing it on the scope?

[Chuck H]

If making changes to the input didn't have ANY affect on the problem I would think that the problem isn't on the input. Even though it's the most sensitive stage in the amp. You could use the scope to trace the oscillation.

Some guitar amps simply can't reproduce a square wave accurately. They're designed as signal processors, basically, so accurate reproduction takes a back seat to listening tests. It does take fairly accurate reproduction to manage a square wave input.

What is the strength of your input signal?

[Enzo]

"With cable plugged in" confuses me. Other end of cable? COnnected to generator? Generator is off? On?


An oscillatory signal in the amp doesnlt mean the amp is oscillating, just SOMETHING is oscillating. The amp could be simply amplifying some HF signal it is coupling to. Such as an oscillator inside the generator. Plug that cable into a guitar instead, and sit it there. Now what?

[Steve Conner]

Here's one subtle ground loop that has bothered me. It once ruined a whole day's test results.

You take your scope probe and hook it up to look at the amp's output. The tip goes to the positive terminal of the dummy load, and the ground clip goes to the negative terminal.

Now you connect the scope to the amp input, either clipping it there directly, or hooking its trigger terminal to your signal generator which is feeding the amp input.

You now have a ground loop: the speaker return current is creating an IR voltage drop in the negative speaker wire back to the power amp, and that voltage appears on your scope chassis, and eventually finds its way back to the amp input. Result instability. I first noticed this when working with high powered solid-state amps, but it is surely possible with tube amps too.

The best fix I found was to leave the scope's ground clip off the negative speaker terminal. Just connect the tip to the positive terminal.

[Enzo]

I started doing that years ago. One clip wire from the scope ground binding post to amp chassis. NO clip on the probe. Especially on amps with bridged outputs like a lot of Crowns.

[km6xz]

Also, what is the cable connected? Terminated, standard coax(RG-58 has 32pf/foot capacitance and needs to be terminated at a lower z), RF sources around such as a florescent mag light(mine has tremendous RF radiation at around 300khz so it has to be turned off for accurate noise measurements even with low gain circuits) etc.

There is a good chance that you have a no problem with the amp.

[jamesmafyew]

Yeah at this point I'm certain I have no problem with the amp.

To be clear, here's how it was all hooked up when oscillating:
5218 sine wave output -> banana to 1/4" shielded cable -> amp input. Signal generator, load=internal, on OR off w/ zero output, same results
5218 square wave output -> banana to bnc cable -> tek 465's external trigger input
amp's speaker output -> 1/4" cable -> homemade 16-ohm dummy load
tek 465 channel A -> normal shielded x10 probe -> tip to + of dummy load, ground clip to - of dummy load

It was definitely oscillation and not something picked up elsewhere. All knobs at ten, it would do this, reduce either the pre volume or the master or the PPIMV, or even the treble control in the tone stack, the oscillation would rapidly just drop to nothing. I went through the amp stage by stage and it looked like it was absolutely the power stage coupling to the input stage, because I couldn't get it to oscillate at anything other than 10 on all knobs (including PPIMV). It never did it with no input to the amp, i.e. input stage grids grounded. I tried ungrounding the banana->1/4" cable, the oscillation got worse. It was when I unplugged the square wave TRIGGER cable to the scope that the oscillation stopped, and no matter what anything else was hooked up to or set to, couldn't get it to do it again.

So there are a number of ground loops going on here, I realize that, I was just wondering why breaking this particular loop (through the trigger input) would cause such a dramatic change. Steve's explanation makes the most sense.